Kevin Kelly on How Generative AI is Changing the Way We Work (SXSW 2023)

Legendary Kevin Kelly gave an excellent talk at SXSW 2023 about his vision of AI and Work. Some ideas from his talk, worth watching in full:

• We only call AI the stuff that AI can't do yet. For things that AI can do, we use specific names: Machine Learning, Large Language Models, etc. (This reminded me of Arthur C. Clarke's idea that any sufficiently advanced technology is indistinguishable from magic.)
• We shouldn't worry about AIs being too intelligent. There are already lot of things or apps that are smarter than we are—take, for example, a calculator. It's the way we enhance our capabilities.
• We have still a poor understanding of how we think. There are many kinds of minds, many kinds of thinking. We should not think about AI (singular) but AIs (plural) with different qualities and different tendencies. We can't generalize by saying "AIs will be smarter than us." They will—already are—smarter than us in many ways, and dumber than us also in many ways.
• We can consider AIs like alien intelligence because they are not like humans, they don't think like us. It's harder for us to think differently when we are all connected together, 24/7. But thinking differently is the engine for innovation. AIs will help us think differently.
• Consciousness for AIs, if we achieve it, will be in most cases a liability. Do you want your AI-driven car to be conscious?
• There is Creativity (uppercase) and creativity (lowercase). AIs are already helping us with "lowercase creativity". This kind of creativity has become a commodity.
• The AI inventors didn't know what their AIs could do. It's the million of people using AI daily the ones who are discovering the uses of AIs.
• Everybody should be using AIs every day, and learning how AIs can help them, learning the quirks of making AIs work for you.
• Productivity is for robots. Productivity is a terrible metric for humans. You have to try things that you can fail at to learn. That's one of the bases of science. The things that we value as humans are inherently inefficient.
• We can work with AIs as someone who works with an intern. You don't present your intern's work "as is" or as yours. Everybody can spot that. But it can be a starting point. It can help overcome blank page syndrome. It can be your research assistant.
• With AIs, we get "wisdom of the crowds" type of knowledge. We get the average. It can be very useful, but it's not necessarily our best. So we have to push the AI. We have to engage in roleplaying with it: "suppose you are an…". Like the human average, AIs can be racist, sexist, mean… but we will not tolerate it. We want AIs to become better than us.
• A real game-changer will be AI tools that allow us in the future to generate complete 3D scenes—movies—and worlds by ourselves. Today you cannot create a whole movie by yourself. But AIs will allow us to do it.